I am a very beginner with array data analysis. I have 3 groups CTR and TREATMENT1 and TREATMENT2. I used a very simple design to perform the 3 comparisons, but I found that 2 out 3 Affymetric probesets for actin beta resulted up-regulated in the CTR condition. All the other probsets for actin beta don't significantly change. Is that normal? could it be a sign of something wrong either during the hybridization or during the analysis? Thanks -- output of sessionInfo(): eset = exprs(expr)[, c(1:18)] designM = model.matrix(~ 0 + factor(c("C", "C", "C", "C", "C", "C", "T1", "T1", "T1", "T1", "T1", "T1", "T2", "T2", "T2", "T2", "T2", "T2"), levels = c("C", "T1", "T2"))) colnames(designM) = c("C", "T1", "T2") fit <- lmFit(eset, designM) contrast.matrix <- makeContrasts(C-T1, C-T2, T1-T2, levels = designM) fit2 <- contrasts.fit(fit, contrast.matrix) fit2 <- eBayes(fit2) -- Sent via the guest posting facility at bioconductor.org.

Hi Michele, On 2/13/2012 7:51 PM, Michele Bellesi [guest] wrote: > I am a very beginner with array data analysis. > I have 3 groups CTR and TREATMENT1 and TREATMENT2. I used a very simple design to perform the 3 comparisons, but I found that 2 out 3 Affymetric probesets for actin beta resulted up-regulated in the CTR condition. All the other probsets for actin beta don't significantly change. > Is that normal? could it be a sign of something wrong either during the hybridization or during the analysis? It's hard to say, particularly without access to the data. The usual assumption is that beta actin is constitutively up-regulated, and never changes expression. I have no idea if this is true, and have always harbored doubts. If these were my data to analyze, I would be looking at the fold change difference (are the differences statistically significant, but not likely to be biologically meaningful?), as well as what the probesets are measuring (is it possible that they are picking up transcript variants?). Anyway, you will likely need to delve deeper into your data to see exactly what these differences really might mean. Best, Jim > Thanks > > -- output of sessionInfo(): > > eset = exprs(expr)[, c(1:18)] > > designM = model.matrix(~ 0 + factor(c("C", "C", "C", "C", "C", "C", "T1", "T1", "T1", "T1", "T1", "T1", "T2", "T2", "T2", "T2", "T2", "T2"), levels = c("C", "T1", "T2"))) > colnames(designM) = c("C", "T1", "T2") > fit<- lmFit(eset, designM) > contrast.matrix<- makeContrasts(C-T1, C-T2, T1-T2, levels = designM) > > fit2<- contrasts.fit(fit, contrast.matrix) > fit2<- eBayes(fit2) > > -- > Sent via the guest posting facility at bioconductor.org. > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- James W. MacDonald, M.S. Biostatistician University of Washington Environmental and Occupational Health Sciences 4225 Roosevelt Way NE, # 100 Seattle WA 98105-6099